Taking Linux Public

The president of VA Linux Systems shares his views on Linux and what the


If Linus Torvalds is the Thomas Edison of Linux, the
inventor, then Larry Augustin wants to be its Henry Ford, the great
mass-market builder. Augustin is the founder and president of VA Linux
Systems (formerly VA Research), the Mountain View, CA company that is a leader in the nascent Linux-based hardware market. A
Stanford University Ph.D., Augustin chummed around with Yahoo’s David
Filo and Jerry Yang before starting VA in 1993. His company has been in
the news a lot lately; it recently obtained the rights to the Linux.com
domain name; the company has been hiring key Linux kernel hackers, and
it is expected to be one of the first Linux companies to go public.
Augustin — one of the great gossips of the Linux world — recently took
time out from a driving schedule to sit down for dinner with Linux
Magazine contributor Lee Gomes. The conversation covered a wide terrain:
the work habits of Linus Torvalds, how VA will survive in a world where
everyone does Linux, the religious debates in the Linux community, and
the characteristics of a certain company named Microsoft. It began,
though, with the 36 year-old Augustin discussing one of the recent
setbacks of the Open Source movement.

Linux Magazine: What can be learned from the Mozilla

Larry Augustin: The basic lesson is that just taking any
software project that’s in trouble and making it Open Source doesn’t
magically fix it. It isn’t a silver bullet.

LM: Was Netscape’s action a sign of strength or weakness?

LA: I think it was a sign of weakness. Netscape was losing the
market.Part of their response to Microsoft was to go Open Source. They
weren’t doing it from a position of leadership. When you’re losing and
you have to do something to win, you’re open to more options. But if
Netscape had been winning and was gaining market share, they wouldn’t
have done it.

LM: You call Linux the “Revenge of the Nerds.” Why?

LA: Traditionally, when you’ve got software, you
think of the software as coming from a company. When you think of a
product like Microsoft Office or Access, it’s Microsoft that releases
it. Well, in the Open Source world, the software doesn’t come from a
company. It comes from programmers, from individuals. Rather than being
a company project, software becomes identified with the individuals who
create it. So a lot more of the credit goes to them.

Of course, really good products are rarely designed from the top
down. A great example of what’s wrong with the top-down design is the
Chevy Vega. Did you ever read Lee Iacocca’s book? He’s got this entire
chapter about the Vega. Apparently, General Motors’ management decided
it was going to have a small car to compete with the Japanese. So
management came in and told the engineers, toldthe
designers, “You’re going to design a car that is this big, weighs this
much, runs this fast, and by God, that’s what it’s going to be.” So then
there was this big push to bring the car out. It’s funny, but the Vega
holds the record for the shortest distance ever traveled on the GM test
track by a prototype.

One of the things that management told the engineers was, “You’re
going to have an aluminum block engine.” Well aluminum block engines
warp very quickly. So to try and fix this, they put cast iron heads on
the aluminum block engine. That just made the whole front end heavier.
As a result, they had to beef up the front end of the car. But it went a
half-mile on the test track and the front end fell off. So then they
started bolting things on.

They ended up with a car that had this really tiny, crappy engine,
that weighed a lot, and that was bigger than any of the competition’s.
It also cost more. So GM said, “What do we do now? We’ll put chrome all
over the damn thing and sell it as a luxury car!” So they put these big
chrome stripes around it and it became this small luxury car. And anyone
who’s ever owned a Vega knows it’s a disaster of a car!

LM: Do you think that this focus on individual engineering
feats might be just a transitional thing right now, since more and more
big companies are moving into Open Source?

LA: Perhaps. But I think it’s a fundamental tenet of
good software that behind any good software project, there’s a very
strong, talented lead engineer. So to the extent that Windows 2000 is in
trouble, it’s probably because the project is so big that there’s not
one person at the top who has the whole thing in his head.

LM: The idea that Windows 2000 will be Microsoft’s Vietnam –
don’t you think that might be just wishful thinking by the Linux

LA: To some extent, we like to pick on Microsoft.
It’s one of those easy targets. Is it as bad as we like to say it is? I
don’t know.

LM: Is Linux as good as you like to say it is?

LA: I think Linux isn’t as far along as people say it
is. Linux has got a lot of benefits, but one of my big concerns is that
with all the hype, people will think Linux is going to solve all their
problems. And when they go and try it, they’ll find it doesn’t get the
job done.

LM: At the enterprise level, or at the user desktop level?

LA: Both. This is what happened to NT. Microsoft
pushed NT as their enterprise-level product, and it wasn’t ready. And so
for three or four years, people have been trying to use NT in the
enterprise, and it hasn’t worked for them. It’s getting a black eye.
Well, what happens if people put Linux on their desktop and find it’s
hard to use? Or if they put it in places where they really need
redundant fail-over, or high availability and Linux doesn’t have those
features? Then they’ll say, ‘Oh, this Linux thing, it doesn’t do the job
for me, it doesn’t work.”

LM: What do you think of Linux kernel hackers?

LA: I’m very envious of those guys. To me, that’s the
ultimate thing: being able to create new things like Linux or Apache; to
have written some piece of software that everyone in the world wants to
use. I used to think I sort of knew what was going on. But I’ve gone to
dinner with Linus, David Miller, and Leonard Zubkoff, and I was
completely out of my league. Over my head.

LM: What did they talk about?

LA: At dinner, Dave Miller was beating up Linus
because of a 64-bit address space on Sparc. Linus had designed the
memory management so that the top 1 MB out of 64 terabytes of address
space couldn’t be used. But Dave needed to figure out a way to use every
bit of the address space — and there was this one page that Linus had
set aside as not being used.Dave had figured out a way to reclaim it! I
did a Ph.D. in computer architecture at Stanford. And I have one or two
patches in the Linux kernel. But these guys were over my head

LM: Maybe I’m missing the point, but why did Linus reserve the
page, and how did Dave reclaim it?

LA: I don’t know. It was beyond me. I’d be sitting
there in these conversations and I’d be completely silent, because this
was completely beyond me.

LM: What do you think of the Free Software versus Open Source
Software debate?

LA: Richard Stallman sees it fundamentally as a
religious and political issue, that software should be free. It’s a
political statement; it’s not an economic statement. I’ve always been of
the economic bent. That is, Open Source Software works better. It
produces better code. It provides more options for the customer. It’s
just better for the consumer, and that’s the reason for using it. Not
because of some political or religious argument.

LM: Do you think that Richard Stallman helps or hurts the

LA: Both. Richard provides a guaranteed barometer.
You always know how Richard’s going to come down on an issue; he’s your
litmus test. On the other hand, he tries to sell things to people based
on his political argument and it just doesn’t carry any weight in the
corporate world. It does not benefit Free Software when he makes
outlandish religious arguments about why things should be free.

LM: Do you think that some companies are trying to take
advantage of the hype surrounding Open Source?

LA: It bothers me when corporationstry to take
advantage of the hype without really being open. I think Apple’s release
of pieces of Mac OS X was a great marketing move that fundamentally
didn’t release any real open technology. It’s people taking advantage of
Open Source, trying to get some benefit by association with the

LM: Does VA Research have any proprietary technology?

LA: Part of the reason I wanted to get into the
hardware and systems business is that the temptation to do proprietary
things with software is much lower. We can do things that are
proprietary around the hardware and the system, while still keeping all
of the software open. That gives us a way to differentiate ourselves
without being as tempted to do proprietary software.

LM: But if you have a proprietary graphics chip, doesn’t that
go against the principles of Open Source?

LA: Well, it wouldn’t be proprietary in the sense
that the interfaces would be closed. When I say proprietary, I mean
doing things that are unique to us, like a clustered system. Now in
order to do that, we had to build some special hardware that connects it
to the system. We’re telling everyone how to work with that program, and
we’re releasing all the software around it. We may have some patents
around the design of that hardware, but everyone can program it.
Everyone can work with it.

LM: Do you have a fundamental moral problem with proprietary

LA: I don’t have a fundamental moral problem with it.
I’ve consulted in places that work on proprietary software. I think that
economically and quality-wise, Open Source works better and I encourage
people, as theywork on projects, to consider opening them up and looking
at building their business models around support. People will pay for
“free” software.


LM: You and Linus are friends. What’s he like?

LA: Linus is a very nice, normal guy. When we have
dinner, we don’t talk business. We talk about family and kids, and the
kids come around and he’s chasing them around and playing with them. And
all I can think is: How does he possibly have time to have a family and
pay attention to his kids and have a job at TransMeta anddo
Linux? Successful people like Linus are extremely efficient. He’s just
smarter than the rest of us.

LM: Are there some things Linus ought to change about how he
manages Linux?

LA: I wish he would make it easier for other people
to contribute. That’s one of the things that worries me about Linux.
Suppose that some brilliant person comes along with a patch for Linux,
and he e-mails it to Linus. Well, what if Linus is on vacation that
week? Next week, he comes back and says, “Oh my god! I’ve got 3,000
messages. I can’t read them all, so I’ll just delete them.” So this
brilliant person who has made this wonderful patch has the patch thrown
away –not because there’s any problem with it, but because Linus, as an
individual, got overwhelmed. And that happens occasionally. Some person
who’s very smart, who could contribute a lot, is going to go away just
because Linus deleted the message.

LM: So what should he change?

LA: The answer to that is to try to find a way to
manage these patches so that Linus is not the only person involved.
There should be one place where everyone sends patches, and maybe three
or four people who Linustrusts that help him go through them. They
wouldn’t be making the final decision, but helping Linus evaluate,
filter, and track.

There are people who are working on this problem. There are plenty
of people in the Open Source community who recognize that Linus’
bandwidth is a potential problem. For example, Larry McVoy is working on a project called BitKeeper, and his
goal is for Linus to use BitKeeper to manage the source code.

LM: What is Linus’ feeling on that?

LA: I think Linus feels that Linux is his project and
that means that nobody else writes to that code.

He’s managed this project that’s gone on what, seven years now? It
has attracted thousands of developers, has a great core set of really
technically competent people, and has captured the imagination of the
world — enough so that you’re willing to sit down here and take
pictures of me eating dinner in Palo Alto.

LM: But do you worry that Linus might not be able to scale up
to the next level?

LA: No. Eric Raymond has this great saying — that the
community created him because Richard Stallman wasn’t filling the next
need of being able to take Open Source to the business community. I
think one of the great things about Open Source is that it has this
natural evolutionary tendency to solve problems. The Open Source model
will find a way.

LM: Why isn’t there a Linus Torvalds Foundation where Linus
works with a staff and a secretary?

LA: I don’t know if he wants that. In some sense,
he’s a modest guy, and this concept of having a staff is strange for
him. But I would fund it. I would like to make it easy for him to spend
more time doing what he does really well. And that is working on the
code and then making a reasonable number of public appearances. He’s a
wonderful spokesman.

LM: What do you think of the way the Linux distributions are

LA: I think that it’s very, very important for Linux
that there be a “Linux market.” There can’t be a “Caldera market,” a
“SuSE market” or a “Red Hat market.” To the extent that those become
different markets, we recreate the Unix problem.

LM: Do you worry about that?

LA: I worry when a software vendor announces that
they are going to port to one distribution and not the others. We’re in
the Linux business, so we want to support things like the Linux
Standards Base and other standardization efforts in such a way that the
whole thing grows and comes together. If a customer calls us up and
says, “I have to have so and so’s distribution because I’m going to run
this software,” that’s a problem for us. And if Oracle or Informix runs
only on certain distributions, that’s a problem for us.

LM: Do you see yourselves as being distribution-agnostic?

LA: We want to be able to be distribution-neutral.
Our default is to send Red Hat to customers. But we want to make sure
that if a customer wants SuSE or Caldera or Debian, that we can install
it and it will work for them.

LM: As Linux becomes more popular, it seems likely that
computer companies will begin to make hardware especially for it. Just
how different is a PC that is tailor-made for Linux from a garden
variety PC?

LA: Very. For example, we build a clustered system
where all the management busses are wired so that you’ll have a single
interface on one screen that will tell you about every fan in every
system and whether or not it’s spinning. If a hard drive fails anywhere
in your cluster of a thousand machines, there’s a little red flashing
spot for that hard drive. That machine is targeted at a specific Linux
market for clusters. We’re doing the same thing for ISPs; so we’ve got a
line of products we’ve built around the ISP market. We’ve chosen
particular hardware options there that we know work well with Linux and
frankly we don’t know how well they work with NT. Some of them may not
work well with NT.

LM: How often do you compete against Sun?

LA: Not as much as you would expect. It’s very strange. Sun is
a company that I would expect us to go head to head against. In fact, we
tell customers, “Gee, why should you buy a Sun box when you can get a
Linux box from us that’s half the price and three times the
performance?” The phrase I’ve heard about Sun’s approach to Linux is
that it is “firmly conflicted.”

LM: Do you ever see yourselves becoming the next Sun?

LA: I think we could be the size of a Sun. I like the Sun
business model, except for the fact that they’re too closed. The idea of
being able to combine hardware and software and produce a better system
because you can control both of them — that’s our fundamental business
model. I think Sun has that idea right.

LM: Let’s say that in a few years, Linux has 50 percent of the
enterprise market. At that point, are you guys a mainstream or a niche

LA: Mainstream.

LM: But why won’t SGI or Dell or Compaq or HP or IBM get that
mainstream business?

LA: Because we’re going to focus on building products for
Linux. I think we’ll succeed at it because we recognize that,
fundamentally, it’s a Linux market, and that Linux is the core of our
business and we do everything to grow around Linux. Whereas those other
companies are adding Linux on as another thing they do around the edge
of their business. Our goal is to be the major market player around
Linux systems.

LM: But if it is a growing market, won’t they do the same

LA: These guys are going to get into it. They’re going to do
it, but we have the opportunity to be the company at the core, the
company that grabs the big chunk of that market.

LM: But why can’t they do Linux too?

LA: I think Linux is fundamentally different from the way they
do business now. This notion of setting up a group of engineers that do
software you don’t own — that’s a different model. It’s also a
different model to create this group of engineers that becomes known for
what it does as opposed to what the company does. In the HP model, HP
does software, but in the Linux model, it’s this engineer who happens to
work for HP who does software.

LM: Red Hat Software just filed for its initial public
offering. Will VA Systems be public in a year from now? Or will it be a
unit of IBM, or something else altogether?

LA: I think we’ll probably be a public company a year from
now. I think a year from now you’re going to see Linux start showing up
in the business desktop. And I think that a year from now we’ll be
releasing some interesting new products. The ability to combine hardware
development with OS development is something few companies have had for
a long time. Apple was able to do some interesting things, but they’ve
been so closed it’s been difficult for them to grow the market. PC
vendors have been so separated from the OS for so long that they’ve
become unimaginative. But we’re going to change that. We will have some
products that are really cool and interesting, and that show what you
can do when you can control the OS a bit and control the hardware a

LM: Whenever a company does an Initial Public Offering, it has
to publish a prospectus that lists all the risks associated with
investing in the company. What do you think that section of the VA
prospectus will say?

LA: It’s a very competitive market around hardware, around
systems, around software. We’re always going to be under a lot of
pressure from all these other companies. Some of them are a lot bigger
than us. I think that is the single biggest risk. That’s pretty

LM: How will you guys deal with the constant commoditization
and falling prices of the computer industry?

LA: It’s good for us, in the sense that we can take advantage
of it. One of the issues I had with Sun 16 years ago, and part of the
reason I started VA Research, was I saw Sun going off in this direction
that didn’t let them leverage the industry trends you describe. But at
VA, while we want to create these great systems, we also want to do so
in a way that leverages the mass market.

LM: Why did you form VA Research?

LA: I formed VA because there was a need. Personally,
I wanted to run Unix on a PC, but I didn’t trust a PC company to make a
Unix workstation. Unix workstations are up 24 hours a day, seven days a
week. Theyoften run as servers. They have thingslike SCSI I/O busses,
and remote management features. PCs usually don’t have those

Another problem was that no one would support anything but DOS or
Windows on their hardware. If you found a hardware problem (and I found
lots of them), you had to reproduce it under DOS/Windows. On top of
this, if you ran Unix on a PC, you’d often get stuck in a finger
pointing match between the software vendor and the hardware vendor. I
hated getting stuck there.

I felt there needed to be a place where people could work on
advancing Linux development. That meant a place where people could work
on Linux full time, with access to the latest state-of-the-art big
systems. For example, we just hired Ted Ts’o, who works on large scale
filesystem support. We were able to give Ted access to a system with 2
Terabytes of disk space. This is something that Ted justwouldn’t be able
to afford on his own.

And I wanted to make cool computers. I’ve always been an engineer at
heart. I would always open up the latest system and begin dissecting it,
looking for ways to make it better.

LM: You’re the “A” in VA Systems, right? Who is the “V?”

LA: The “V” was James Vera. When the company became a
full-time business James decided to stay at Stanford. We kept the name,

LM: What’s the story with you and Yahoo? Is it true that you
turned downa chance to get in on the ground floor?

LA: Dave Filo, Jerry Yang, David Ku (who was at
Escalade last I heard), and I wrote Internet business plans together.
When Yahoo began, I had already started the Linux computer business and
was selling systems. Yahoo was running off of Jerry’s DEC workstation in
his office at Stanford. We all knew the Internet would be big, and we
had this vague notion that there was a business in it somewhere. At the
time, I don’t think they really had any idea how Yahoo could make money.
We floated around several business plans for a while, and finally we
ended up following separate paths. I continued with Linux, and they
continued with Yahoo. Certainly they ramped sooner, but we’re catching

LM: What do you think of the state of software innovation in
Linux? It seems that a lot of people are proud of the fact that they
have taken Linux and made it look like Windows.

LA: You’re looking at the wrong user interface. KDE
is a great example of making Linux look like Windows. Youshould look at
some of the stuff that’s starting to come out of the Enlightenment
project. Gregg Zehr, my VP of engineering, comes from Apple, and he and
I started talking Gnome and Enlightenment, and he said, “Yeah, yeah,
yeah. User interfaces are old. We knew how to do them at Apple. It isn’t
going to change.” But then he sat down with Mandrake, and Mandrake
started showing some of the things Enlightenment could do. And Zehr
said, “We’re going to clean this up, and a year from now, I’m going back
over to Apple, and they’re going to be so darn envious.”

LM: Who is Mandrake?

LA: Geoff Harrison. He is the co-author of the Enlightenment
Window Manager, which is the default window manager on GNOME. They’ve
created this really fast, lean window manager that lets people create
these sort of arbitrary back-decorations on windows.

For example, you can have a completely transparent background, with
floating text hanging anywhere in the middle of the screen. Now that in
itself sounds strange, but the point is they are all creating new ways
of visualizing work and new ways of interacting with the user

LM: We were talking earlier about the business desktop. How
much do you think business people want to learn about computers?

LA: Gregg Zehr’s motto is, “One button, no instructions.” And
that’s what I think business desktop users want.

LM: If the Open Source model is so good, why hasn’t it come up
with the next Microsoft Office?

LA: Give us time!

LM: It’s been years.

LA: We just got the operating system! Look, we did the
compilers, we did the utilities, and then it took us six years to get
the operating system. It could take us another five years to have an
Office suite. It’s a hard problem and it isn’t going to happen
overnight, but people are working on it.

LM: So why not just use Word on Windows?

LA: People run Windows because it runs Word. They don’t run
Word because it works on Windows. People don’t want to run Windows; it’s
Windows that most people have problems with. If people could, they would
be happy to run Word on Linux. I think that there are bugs in Word that
people would like to fix. If it were Open Source, it would be better.
But right now if Word were available on Linux people would run it.

LM: What do you think would happen if Microsoft made Office
available on Linux?

LA: The business desktop would go to Linux immediately. I’m
not sure about the home desktop. Business people want to have a very
small number of system administrators who tightly control what’s
available to users. They don’t want users to be able to install their
own software; they don’t want users to be able to break the systems;
they want users to be constrained within their own space, and they want
all of that to be remotely managed by a system administrators, sitting
in — you know –Tulsa, managing machines in New York and San

LM: What do you think Microsoft’s stock price will be say two
years from now?

LA: I think Microsoft’s stock will continue to do well.
Microsoft is a big company. They have this dominant position in the
market and Linux is eating away it. But Microsoft is such a big company
that it’s going to be more than three years before its stock takes a
hit. Look at IBM. IBM went through this period when it was the most
profitable company in America. But five years later? Its stockprice went
through the floor! But IBMre-invented itself, and now look at its stock

LM: So what would a re-invented Microsoft look like?

LA: Microsoft is going to face this issue down the road as
Open Source becomes more prevalent. Maybe they will do more Open Source.
Maybe they will turn into the leading Linux applications vendor.
If Microsoft doesn’t re-invent itself at some point, it will go the way
of Apple. The iMac is a great product, but it’s not a re-invention of
how Apple does business. One big problem for Microsoft is that they are
beginning to run into a serious software scaling problem. That’s one
reason they can make a really good browser but not a really good
operating system. The thing about software design is it’s so easy to
create more code. In hardware design, adding transistors is actually
hard, so there is a self-limiting factor in the design. But in the case
of software, anybody can spew out thousands of lines of code that are

LM: Isn’t that a problem for Linux as well? Doesn’t Linus want world domination?

LA: Oh! But world domination is not the same as making a
single piece of software that does everything. Linus is just making a
little tiny piece of software that other people can build around, and
the end result of people building on this piece is world domination.
See, I think Microsoft has this tendency to pull everything in around
the operating system. It’s a fundamental business control model. It’s at
the core of their business.

LM: Do you think Linux will be competing with Windows CE in
the embedded device market?

LA: Yes. Linux is something that a lot of embedded software
people are looking at. When you’re looking at a $100 or $200 device, a
$20 operating system is hardly insignificant. Plus, people worry about
being tied to one vendor. I think the PalmPilot a is great example of
how a device can succeed without necessarily having to be based on
Microsoft. I think that it could inspire a lot of people to go to

LM: Does it upset you that AT&T got a $5 billion
investment from Microsoft as long as AT&T used Windows CE in its
set-top boxes. Isn’t that an example of how Microsoft has to pay
people to use its software?

LA: Very rarely has Microsoft’s software succeeded on
technical merit. Oh, they did some very good applications. But look at
Microsoft; they have succeeded at least as much on marketing and on
buying into markets.

LM: Do you think there is anything stopping Microsoft from
distributing Microsoft Linux?

LA: They can’t do a Linux distribution. We win when
they do that.

LM: Do we win? Or does it just shift the market away from the
existing Linux players into Microsoft’s marketing machine?

LA: Well, a Microsoft Linux would have to be fundamentally
Open Source. They have to give back their changes, their enhancements.
Let them make it better.

LM: The chance of them doing that?

LA: Zero.

Lee Gomes covers technology for the Wall Street

Comments are closed.